Human Coherence Research
Most modern technology runs on a simple loop: show something, measure what you click, then show you more of what keeps you clicking.
That is not inherently evil. It is a sane strategy when the only thing you can reliably observe is external behavior. But it also produces a predictable outcome: products compete to capture attention and shape action, not to improve the quality of a person's inner life over time.
AI makes a different move possible. For the first time, systems can start incorporating internal context—signals that reflect how an experience lands in your nervous system, your attention, your emotional dynamics. If we can sense internal context, we can build choice architecture oriented around coherence rather than conversion.
This essay opens a research project. Not a manifesto. Not an investment pitch. A genuine attempt to figure out whether this idea has substance.
What I Mean by Coherence
Coherence is a working construct, not a single settled variable. In plain language, it points to patterns of internal organization that make a person more stable, less fragmented, and more able to act intentionally. It is something you can sometimes feel and sometimes measure indirectly. It becomes a goal only when someone explicitly chooses it.
To keep this practical, I separate coherence into layers that can be studied before we pretend we know what we're doing:
- Physiological regulation: Autonomic stability, recovery capacity, resilience to stressors.
- Cognitive coherence: Attention stability, reduced fragmentation, the ability to sustain focus without compulsive switching.
- Affective coherence: Emotion dynamics like recovery speed and flexibility, not the absence of negative emotion.
- Values coherence: Behavior matches stated aims; tradeoffs are chosen rather than drifted into.
A boundary worth stating early: coherence is not moral superiority, perpetual calm, productivity at all costs, or compliance with an external agenda.
The Basic Hypothesis
Today, most choice architecture optimizes external stimuli to drive external behavior. AI allows the loop to include internal context. Not mind reading. Something more modest and more testable:
If we can measure proxies of internal state over time, then we can build tools that help people notice patterns, form better hypotheses about what supports them, and make better choices.
Why This Might Be a Real Category
Three forces are converging.
First, measurement is no longer scarce. Wearables and phones produce continuous streams of physiological, behavioral, and context data. Apple, Oura, Whoop, Garmin, and others have made sensing normal. The bottleneck is not inputs. The bottleneck is interpretation and responsible influence.
Second, foundation models are finally capable of integrating messy context. Humans live in noisy, confounded reality. We drink coffee, sleep badly, travel, argue, exercise—then try to infer what any one signal "means." That is exactly the kind of ambiguity that modern AI is built to handle, at least probabilistically.
Third, the costs of the attention economy are becoming visible. People feel scattered. Teams feel brittle. Culture feels reactive. Whether technology can help without becoming surveillance or manipulation is an open question, but the demand signal is clear.
So the project is not "coherence is the future." The project is: does a coherence-oriented architecture produce products that people want, trust, and pay for while staying inside acceptable ethical constraints?
Why I'm Doing This
I have spent most of my career watching how narratives organize people in startups, in families, in institutions. I've built a company around trust and data. I've worked with hundreds of founders through StartX. I've written about meaning-making and consciousness for years.
The coherence economy, if it exists, lives where these threads converge.
What Comes Next
Over the coming months, I'll be mapping the landscape of companies and labs working in this space. I'll be studying what's technically feasible in measurement, interpretation, and influence. I'll be talking to researchers, founders, and practitioners. And I'll be publishing as I go to accelerate engaging with the right people.
What's Next: The taxonomy I'm using to study this space. A framework for distinguishing what is measured from what is inferred from what is changed.
Brendan Marshall
Back to research